update to latest hdt commit, error out if not triples found, additional w3c testing

Author: GregHanson

Cargo.lock

@@ -15,6 +15,7 @@ name = "benchmark"
 harness = false
 
 [dependencies]
+anyhow = "1.0"
 chrono = "0.4"
 clap = { version = "4.5", features = ["derive","cargo"] }
 clap-verbosity-flag = "3.0"

Cargo.toml

@@ -364,6 +364,19 @@ mod tests {
 
             let res = build_hdt(vec![f.clone()], new_hdt.as_str(), Options::default());
             assert!(res.is_ok());
+
+            let source = std::fs::File::open(new_hdt).expect("failed to open hdt file");
+            let hdt_reader = BufReader::new(source);
+            let res = Hdt::new(hdt_reader);
+
+            match &res {
+                Ok(_) => {}
+                Err(e) => {
+                    eprintln!("{f} failed to load constructed hdt: {e}")
+                }
+            }
+
+            assert!(res.is_ok());
         }
     }
 
@@ -387,7 +400,25 @@ mod tests {
             match &res {
                 Ok(_) => {}
                 Err(e) => {
-                    eprintln!("{f} failed to convert: {e}")
+                    // files with no triples should error out, the HDT library errors out as well
+                    if !f.contains("empty.ttl") {
+                        eprintln!("{f} failed to convert: {e}");
+                    }
+                    assert!(f.contains("empty.ttl"));
+                    continue;
+                }
+            }
+
+            assert!(res.is_ok());
+
+            let source = std::fs::File::open(new_hdt).expect("failed to open hdt file");
+            let hdt_reader = BufReader::new(source);
+            let res = Hdt::new(hdt_reader);
+
+            match &res {
+                Ok(_) => {}
+                Err(e) => {
+                    eprintln!("{f} failed to load constructed hdt: {e}")
                 }
             }
 
@@ -420,6 +451,19 @@ mod tests {
             }
 
             assert!(res.is_ok());
+
+            let source = std::fs::File::open(new_hdt).expect("failed to open hdt file");
+            let hdt_reader = BufReader::new(source);
+            let res = Hdt::new(hdt_reader);
+
+            match &res {
+                Ok(_) => {}
+                Err(e) => {
+                    eprintln!("{f} failed to load constructed hdt: {e}")
+                }
+            }
+
+            assert!(res.is_ok());
         }
     }
 

src/builder.rs

@@ -5,7 +5,7 @@ use hdt::{
     containers::{self, ControlType, Sequence, vbyte::encode_vbyte},
     dict_sect_pfc::DictSectPFC,
 };
-use log::{debug, error, warn};
+use log::{debug, error};
 use oxrdf::Term;
 use oxrdfio::RdfFormat::NTriples;
 use oxrdfio::RdfParser;
@@ -90,7 +90,8 @@ impl FourSectDictBuilder {
             object_terms.insert(term_to_hdt_bgp_str(&q.object)?);
         }
         if dict.predicate_terms.is_empty() {
-            warn!("no triples found in provided RDF");
+            error!("no triples found in provided RDF");
+            return Err(anyhow::anyhow!("no triples found in the provided RDF file").into());
         }
 
         dict.shared_terms = subject_terms.intersection(&object_terms).cloned().collect();